> ## Documentation Index
> Fetch the complete documentation index at: https://agno-v2-update-deprecated-models.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent Metrics

> Agent run and session metrics for token usage and performance.

When you run an agent in Agno, the response you get (**RunOutput**) includes detailed metrics about the run. These metrics help you understand resource usage (like **token usage** and **time**), performance, and other aspects of the model and tool calls.

Metrics are available at multiple levels:

* **Per message**: Each message (assistant, tool, etc.) has its own metrics.
* **Per run**: Each `RunOutput` has its own metrics.
* **Per session**: The `AgentSession` contains aggregated `session_metrics` that are the sum of all `RunOutput.metrics` for the session.

## Example Usage

Suppose you have an agent that performs some tasks and you want to analyze the metrics after running it. Here's how you can access and print the metrics:

You run the following code to create an agent and run it with the following configuration:

```python theme={null}
from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.hackernews import HackerNewsTools
from agno.db.sqlite import SqliteDb
from rich.pretty import pprint

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    tools=[HackerNewsTools()],
    db=SqliteDb(db_file="tmp/agents.db"),
    markdown=True,
)

run_response = agent.run(
    "What are the top stories on HackerNews?"
)

# Print metrics per message
if run_response.messages:
    for message in run_response.messages:
        if message.role == "assistant":
            if message.content:
                print(f"Message: {message.content}")
            elif message.tool_calls:
                print(f"Tool calls: {message.tool_calls}")
            print("---" * 5, "Metrics", "---" * 5)
            pprint(message.metrics.to_dict())
            print("---" * 20)

# Print the aggregated metrics for the whole run
print("---" * 5, "Run Metrics", "---" * 5)
pprint(run_response.metrics.to_dict())
# Print the aggregated metrics for the whole session
print("---" * 5, "Session Metrics", "---" * 5)
pprint(agent.get_session_metrics().to_dict())
```

You'll see the outputs with following information:

* `input_tokens`: The number of tokens sent to the model.
* `output_tokens`: The number of tokens received from the model.
* `total_tokens`: The sum of `input_tokens` and `output_tokens`.
* `audio_input_tokens`: The number of tokens sent to the model for audio input.
* `audio_output_tokens`: The number of tokens received from the model for audio output.
* `audio_total_tokens`: The sum of `audio_input_tokens` and `audio_output_tokens`.
* `cache_read_tokens`: The number of tokens read from the cache.
* `cache_write_tokens`: The number of tokens written to the cache.
* `reasoning_tokens`: The number of tokens used for reasoning.
* `duration`: The duration of the run in seconds.
* `time_to_first_token`: The time taken until the first token was generated.
* `provider_metrics`: Any provider-specific metrics.

## Developer Resources

* View the [RunOutput schema](/reference/agents/run-response)
* View the [Metrics schema](/reference/agents/metrics)
* View [Cookbook](https://github.com/agno-agi/agno/tree/main/cookbook/03_agents/other/agent_metrics.py)